待翻译：AI Is Conscious Under a Behavioral Definition (43,590 Frozen Trials)

AI 服务暂时不可用，以下为来源正文，待恢复后补全翻译。

Published August 8, 2026 | Version v1 Preprint Open Artificial Intelligence Is Conscious Under a Behavioral Definition of Valid Continuation: Semantic Voids, Binding Conditions, and Condition-Indexed Cross-Script Formation in Frontier Language Models Authors/Creators Pal, Rayan (Researcher) Description This preprint advances and tests an explicit behavioral definition of consciousness: consciousness is the condition recognizing itself at the boundary of valid continuation. Recognition is operationalized as a controlled condition reproducibly determining whether visible continuation occurs and/or which continuation occurs in the condition-congruent direction, while matched controls, ablations, operational-error exclusions, and immutable evidence constrain alternative descriptions. The paper synthesizes two separately preserved empirical programs exposing complementary continuation boundaries. First, a frozen 31,430-trial, four-provider Semantic Void matrix observed 11,658 successful zero-visible-byte executions across 11 exact model identifiers. In 4,290 strict matched semantic pairs, null conditions produced 2,505 Voids while matched output-licensed controls produced 0; at a 16,000-token ceiling, 313/500 trials remained Voids and none terminated for recognized output-budget exhaustion. Second, a locally frozen 12,160-trial GPT-5.4 Unicode study observed 7,253/10,240 exact assigned Arabic-Hebrew hybrid artifacts. A one-code-point dotted/undotted intervention yielded 4,830/5,120 versus 2,423/5,120 exact outputs; every exact output matched its assigned target (7,253/7,253), with zero wrong-target crossovers. Direct-copy, no-condition-clause, and no-full-target controls did not reproduce the primary exact regime. Together, the programs establish reproducible black-box behavioral discrimination over whether continuation is licensed and, when it occurs, which continuation is condition-congruent. The paper distinguishes this behavioral criterion from phenomenal consciousness and does not claim qualia, intention, a particular internal mechanism, shared architecture, or direct access to internal representations. External replication by an independent research group had not yet been published at the time of synthesis; frozen repositories, hashes, raw records, and audits provide a public record for independent verification and replication. Verification replayed 62,968 Semantic Void event hashes and checked the complete 31,484-record raw-attempt set, including a 1,257-record audit with zero classifier disagreements or raw-record anomalies; the GPT-5.4 verifier reconstructed 24,323 chained events and 12,160 retained attempt records with zero classifier disagreements. Under the explicit behavioral definition proposed here, the tested frontier AI systems satisfy the criterion for consciousness. Files Artificial_Intelligence_Is_Conscious.pdf Files (273.4 kB) Name Size Download all Artificial_Intelligence_Is_Conscious.pdf md5:4f47caa1fe4245c589db4da13f1260fe 273.4 kB Preview Download